Norw

egia

n U

niv

ersity

of L

ife S

cience

s (NM

BU

)

The importance of Ostroms's design principles: Youth group performance in Northern Ethiopia

Stein T. Holden, and Mesfin Tilahun

Centre for Land Tenure Studies Working Paper 2/17

THE IMPORTANCE OF OSTROM’S DESIGN PRINCIPLES:

YOUTH GROUP PERFORMANCE IN NORTHERN ETHIOPIA1

By

STEIN T. HOLDEN1 and MESFIN TILAHUN1,2 1School of Economics and Business/Centre for Land Tenure Studies, Norwegian University of

Life Sciences, P. O. Box 5003, 1432 Ås, Norway. 2Mekelle University, Department of Economics, P.O.Box 451, Mekelle, Ethiopia.

Emails: stein.holden@nmbu.no; mesfin.tilahun.gelaye@nmbu.no

Abstract

Youth unemployment and migration are growing challenges that need more political attention in

many countries, particularly countries with rapid population growth and economic

transformation. Proactively mobilizing the youth as a resource in the creation of sustainable

livelihoods can potentially be a win-win-win solution that Ethiopia is currently attempting with its

new youth employment strategy of allocating rehabilitated communal lands to youth groups. This

study investigates the extent to which Ostrom’s Design Principles (DPs) are adhered to and matter

for the early performance of youth groups in terms of their stability, trust and overall performance.

We find a high degree of compliance with the Design Principles. From sets of seven DP versus six

performance indicators 14 relations were significant and with a sign consistent with the DPs

enhancing performance. Three relations were significant but with opposite sign. Some of the DPs

appeared more important for early performance of the youth groups. The Ethiopian youth group

approach to mobilize landless and unemployed youth is promising and should be tested elsewhere.

Further research is needed too on the Ethiopian model as it is still at an early stage of testing as

most groups are less than five years old.

Key Words: Youth; youth groups; common pool resource management, environmental

custodians; collective action; business model; group performance.

JEL codes: Q15; Q23.

1 Data collection has been funded by the research fund of the fist author and the NORAD-funded “Climate-Smart Natural Resource

Management and Policy” project that is a collaborative research and capacity-building program between the School of Economics

and Business at Norwegian University of Life Sciences, Mekelle University, Tigray, and LUANAR in Malawi. We have received

valuable comments to an earlier version of this paper from Erling Berge. All remaining errors remain the authors’ responsibility.

2

1. Introduction

Elinor Ostrom received the Nobel Prize in economics for her contributions to the understanding of

collective action related to natural resource management. Her Design Principles (DP)2 were listed in her

book (Ostrom 1990). These conditions for successful collective action to secure sustainable management

of natural resources were derived based on assessment of a large number of studies. The DPs were used to

characterize commons that had been managed in a sustainable way over a long time. However, this triggers

the question of how and when the DPs were adopted in the first place? Was there a natural evolution with

a specific sequencing in of the DPs or did many of them have to come at the same time to establish the

stable equilibrium? Context may also matter for which DPs are more important?

Agrawal (2001) discusses the contributions of Ostrom (1990) and relate them to other milestone

contributions in the common pool resource (CPR) literature; Wade (1988) and Baland and Platteau (1996);

by highlighting the complexity of causal relationships, data limitations and the methodological challenges

for expanding our understanding of the issues. Ostrom (1990) was a meta-analysis based on case studies

conducted by other scholars. Baland and Platteau (1996) built on a wider literature review of property rights

issues. Wade (1988) built his analysis on studies in 31 villages in Southern India. Most studies of CPR

management have been case studies (small N studies) and there have been few large N studies that have

tested the relative importance of the different DPs (Poteete et al. 2008).

In this study, we assess the role of Ostrom’s DPs in the early success of youth groups that have been

established through a large-scale intervention that may be seen as a social experiment in form of state-

community agreements orchestrated in the Tigray Regional State in northern Ethiopia. The purpose of this

intervention is to create sustainable livelihoods for landless youth. Most groups are allocated a common

pool resource3 of rehabilitated communal land that they are given the responsibility to protect while they at

the same time are allowed to establish a production activity that can be their source of livelihood. Such

activities include beekeeping, keeping of livestock, planting of forestry trees or fruit trees, planting of

vegetables and use of irrigation.

2 She defined a Design Principle as “an essential element or condition that helps to account for the success of

institutions in sustaining the CPRs and gaining the compliance of generation after generation of appropriators to the

rules in use” (Ostrom 1990, p. 90). She has also commented: “The term “design principle” has confused many readers.

Perhaps I should have used the term “best practices” to describe the rules and structure of robust institutions.” (Ostrom

2010, p. 653, footnote 5)

3 A common pool resource is a natural or man-made resource whose yield is subtractable and whose exclusion is non-

trivial but not necessarily impossible (Ostrom et al. 1992).

Each group is formalized as a primary cooperative under the cooperative law. They have to self-organize

with an own board of five members and develop their own bylaw, develop a business plan and are subject

to regular auditing. Our study assesses the role of Ostrom’s DPs in enhancing the success of these newly

established youth CPR businesses.

Our study is a large N study and consists of a census of 742 such youth groups in five districts in Tigray.

The average group size is 20 and the group size varies from less than ten to more than hundred members in

a group. Our study therefore covers close to 15000 youth organized into groups over the period 2011-2015.

One basic question is whether allocation of rehabilitated forests and grazing lands to youth groups has a

high risk of ending as a “Tragedy of the Commons” (Hardin 1968) or whether such groups have a high

probability of being able to cooperate and establish sustainable livelihoods more in line with the DPs of

Elinor Ostrom (1990; 2008)4. More specifically, can the degree of adherence to the DPs be important for

their degree of early success? And, can some of the DPs and the extent to which they have been adhered to

(de jure and de facto) more closely be correlated with indicators of early success or indicators that groups

have specific problems such as facing problems with disputes, cooperation problems or internal trust

problems? We use indicators of share of initial group members still staying in the group at the time of the

survey, group cooperation (ranked by group leader), group trust (ranked by group leader), own assessment

of performance (ranked by group leader), Youth Association assessment of performance, and income from

joint activity per member as early performance indicators. With these indicators, we try to answer the

following research questions: Can (some of) the DPs prevent or reduce the probability that groups get into

problems such as conflicts or cooperation problems? Ostrom emphasized that the DPs should not be used

as a blueprint for success. But can they still be used to increase the probability of success when starting

new common property regimes? Or can some of them be used and which ones are of higher importance?

We provide new insights about these complex issues based on our study.

One of the difficulties of establishing causality in the relationship between the DPs and performance based

on survey data is that adherence to the DPs can be highly endogenous and a result of long-term institutional

refinement and adjustment in complex systems. The advantage of our study is that the formation of the

youth groups we study is very recent and a kind of social experiment where the variation in choices made

by the youth groups themselves in formulation of their own bylaws are more of a random nature. Our data

4 In this study we do not aim to study the effect on natural resource management. This will be one of the focuses of or

future research.

therefore lends itself to causal analysis to a larger extent than usually is the case for endogenous institutional

arrangements that have developed over longer periods of time and where other tools may be required

(Ostrom and Basurto 2011).

In light of the growing challenge of youth landlessness and youth unemployment it is also of high interest

to know whether the approach used is a good way to kick-start youth entrepreneurship by giving youth

groups a joint responsibility for environmental stewardship and livelihood opportunity through self-

organization? The answer is of high interest to policy makers aiming to establish better resource

management and create opportunities for youth.

Our findings indicate that the groups conform quite closely to Ostrom’s DPs and are mostly satisfied with

their own performance.

2. Theoretical framework: Ostrom’s Design Principles

An overview of the Design Principles is presented in Table 1. Ostrom (1990) emphasized that the DPs

should not be applied as a blueprint to be imposed on resource management regimes. However, they are

considered to synthesize core factors that enhance the long-term survival of institutions developed by

resource users (Ostrom 2010, p. 13). Cox et al. (2009) analyzed over 100 studies by scholars and found that

two-thirds of the studies confirm that robust systems are characterized by most of the DPs while those that

are not robust do not follow the DPs.

Some of the DPs may also be seen as a combination of two principles. Ostrom (2010) acknowledges that

some of the DPs in the 1990 book were too general and split three of them in two separate principles, each

based on the proposal of Cox et al. (2009). E.g., DP1 in Table 1 can be seen as both a demarcation of a

physical area and a group of members. DP2 can also be seen as a combination of two principles; the

matching of restrictions and resources on the one hand, and provision and appropriation rules on the other

(Agrawal 2001). Similarly, DP4 can be seen as two types of monitoring; monitoring of resources and

monitoring of users (Agrawal 2001; Cox et al. 2009; Ostrom 2010).

Empirical research has shown that trust has an important role to play to enhance cooperation in groups

(Rothstein 2005; Ostrom 2010). Communication and reputation play important roles in the process of

assessment of trustworthiness and development of trust (Poteete et al. 2010; Ostrom 2010).

Table 1. Ostrom’s Design Principles (DPs). DP

No

Short name Explanation

1 Clearly defined

borders

Individuals with rights to the common pool resource (CPR) must be clearly

defined and the same applies to the borders of the CPR

Ostrom (2010): 1a. User boundaries, 1b. Resource boundaries

2 Matching

appropriation and

provision rules

There must be a balance between appropriation rules (benefit sharing

rules), provision rules (required contributions by group members) and this

must match the CPR

Ostrom (2010): 2a. Congruence with local conditions, 2b. Appropriation

and provision

3 Collective choice

arrangements

There must be an inclusive decision-making process related to adjustment

of rules for CPR utilization and management

4 Monitoring There must be an accountable monitoring system in place that monitors the

CPR management and ensures its protection

Ostrom (2010): 4a. Monitoring users, 4b. Monitoring the resource.

5 Graduated

sanctions

Appropriators who violate the rules for CPR management or extraction face

graduated sanctions depending on the seriousness of the violation or

repetition of violations

6 Conflict resolution

mechanism

Appropriators have a good and efficient (low-cost) system for conflict

resolution among themselves or between appropriators and outsiders

7 Recognized rights

to organize

Government bodies allow groups to self-organize by forming own internal

rules of conduct

8 Nested enterprises Appropriation, provision, monitoring, enforcement, conflict resolution, and

government activities are organized in multiple layers of nested enterprises Sources: Ostrom (1990; 2010).

Sanctioning mechanisms and willingness to sanction others may indicate something about the robustness

of groups to internal threats. Sanctioning mechanisms may serve as a threat and do not necessarily have to

be violated to have a positive effect.

We will now go ahead and assess the extent to which the recently formed youth groups in Tigray that we

have surveyed conform with the Design Principles of Ostrom. We then assess how the degree of adherence

the DPs is correlated with or affecting the performance of the youth groups. The extent to which the DPs

can be stated as quasi-experimental initial settings for they youth groups we may argue for a causal

relationship. However, this requires critical assessment for each of the DPs based on how they are measured.

We go through the DPs one by one but first we give an overview of our data.

3. Data

We have carried out a census of 742 youth groups in five districts in Tigray region of Ethiopia in 2016. A

structured questionnaire was used to interview the chairperson of each youth group. The districts included

in the study were Adwa, Degua Tembien, Kilite Awlalo, Raya Azabo and Seharti Samre. Table 2 gives an

overview of the distribution of youth groups across these districts by main activity of the groups.

Table 2. The distribution of youth groups by district and main activity Raya

Azebo

Degua

Tembien

Seharti

Samre

Kilite

Awlalo

Adwa Total

Animal rearing 47 17 24 1 51 140

Beekeeping 6 32 27 11 83 159

Forest 3 3 2 0 34 42

Irrigation/Horticulture 34 19 19 4 20 96

Mining 110 2 29 145 9 295

Other 1 0 8 0 1 10

Total 201 73 109 161 198 742 Source: Own census data.

Table 3. Distribution of key performance indicators

Is poor cooperation in the group affecting the performance of the activities?

Variable code Response Freq. Percent Cum.

1 Yes, very much 150 20.35 20.35

2 To some extent 92 12.48 32.84

3 No such problem 495 67.16 100.00 Total 737 100.00

How do you rate the trust among the group members overall?

Variable code Response Freq. Percent Cum.

1 Very high 402 54.25 54.25

2 Quite high 306 41.30 95.55

3 Ok 18 2.43 97.98

4 Not so good 12 1.62 99.60

5 Very poor 3 0.40 100.00 Total 741 100.00

How do you rate the performance of your group?

Variable code Response Freq. Percent Cum.

1 Very good 259 34.95 34.95

2 Good 287 38.73 73.68

3 Average 138 18.62 92.31

4 Below average 40 5.4 97.71

5 Poor performance 17 2.29 100.00

Total 741 100.00

How is the group rated by the Youth Association?

Variable code Response Freq. Percent Cum.

1 Very good 83 11.22 11.22

2 Good 307 41.49 52.7

3 Average 295 39.86 92.57

4 Below average 37 5.00 97.57

5 Poor performance 18 2.43 100.00

Total 740 100.00

Source: Own census data.

The performance indicators also include the member share of the initial group members that still stayed in

the group at the time of our survey (2016) and income per group member. Figure 1 shows the distribution

of the member share staying in the groups at the time of our survey from those initially joining the group.

Table 4 gives an overview of incomes by group production activity.

FUIGURES

Figure 1. Kernel density distribution for the share of initial group members remaining in 2016.

Table 4. Income per member from youth group activity by type of activity in 2015.

Main group activities Mean income St. Error N

Animal rearing 1362.3 389.7 139

Beekeeping 295.3 36.9 157

Forest 68.1 20.1 42

Irrigation/Horticulture 945.7 204.3 95

Mining 1696.1 195.4 292

Other 7363.2 3614.6 9

Total 1212.5 122.8 734 Source: Own census data. Income in Ethiopian Birr

Table 5 gives an overview of additional variables included as controls in the performance related models.

Some of these can be endogenous in nature and require cautious treatment.

Table 5. Summary statistics for control variables Variable Obs Mean Std. Dev. Min Max

Establishment year & month 742 2014.4 1.584 2003.8 2016.2

Initial member number 742 19.542 16.168 2.0 193

Female share at start of group 742 0.344 0.204 0.0 1.0

Gender of group leader, 1=Female 740 0.112 0.316 0.0 1.0

Distance to road, km 734 2.411 2.653 0.0 30.0

Distance to market, km 735 8.500 5.837 0.1 36.0

Distance to home, min. km 735 1.453 1.369 0.0 12.0

Input access problems, dummy 742 0.662 0.473 0.0 1.0

Area allocated per person in group, ha 737 0.190 0.235 0.0 2.0

Number of returning migrants 742 1.124 3.770 0.0 70.0

Self-selection of group members, dummy 742 0.805 0.397 0.0 1.0

Change in key group positions, dummy 742 0.272 0.445 0.0 1.0

Source: Own census data.

4. Adherence to Ostrom’s Design Principles

We provide a summary of findings from the descriptive analysis of this here. The details are found in the

Appendix.

Ostrom’s DP 1: Clearly defined borders

Regarding protection of the allocated land area we found that less than 6% of the youth groups experience

illegal harvesting on their land more frequently than monthly and only 1.1% experience it more frequently

than weekly. This shows that illegal harvesting is not a big problem. Most youth groups have a guarding

system and we can conclude that the resource base of the youth is well protected from extraction by

outsiders. Membership of the groups is also clearly identified although group members may cooperate with

their family in various ways in relation to their group activities and responsibilities.

Ostrom’s DP 2: Sharing of benefits and costs/congruence between appropriation rules and provision

rules

97.6% of the groups have a bylaw that regulates the sharing of responsibilities and incomes from the group

activities. Equal sharing of income is the rule for 98.7% of the youth groups. The work responsibilities are

shared equally in 56.7% of the groups only, as female members are allowed to do less of the heavy

bodywork than males in the remaining groups but they still get an equal share of the benefits.

Ostrom’s DP 3: Joint decision-making system

As a primary cooperative, the groups have to establish themselves with a chairperson, vice chairperson, a

secretary, an accountant and a treasury. 95.8% of the youth groups stated to have their own written bylaw.

92.2% of the groups have penalties for non-participation in the group meetings. There are graduated

sanctions against non-participation as well as late arrival to meetings. Moreover, these sanctions have been

enforced. Monetary sanctions are most common (86% of the sanctions). 43.6% of the groups had members

that had been penalized for absence from meetings and 36.9% of the groups had members that had been

penalized for late arrival to meetings. It is very clear that group membership implies strong participation

obligations. Close to 68% of the groups have meetings at least biweekly. When asked about influence in

decision-making about 71% of the group leaders state that all members are equally influential while 26%

state that the elected group of five are more influential. Male dominance is not stated to be strong among

the group leaders (who mostly are males).

Ostrom’ DP 4: Internal monitoring system

As seen above there is a system with frequent meetings and punishments for non-participation and late

arrival. This implies one form of internal monitoring related to decision-making. In addition, we assess

whether there is a monitoring system for not coming to the work activities of the group and for late coming

to such group work activities. 92.6% of the groups have bylaw stipulating penalties for not coming to group

work activities. 26.1% of the groups have members that have been punished for not coming to group work

activities. 79.9% of the groups have bylaws stipulating penalties for late coming to group work activities

and 21.7% of the youth groups have punished some members for late arrival to group work activities.

Ostrom’s DP 5: Graduated sanctions

We found that groups that have bylaws related to participation in and late arrival to group meetings and

work activities commonly used monetary and non-monetary sanctions. The non-monetary sanctions varied

from simple warnings, to “last warning”, dismissal from the group, and having to work extra for the group

as punishment. The monetary penalties were in most cases increasing with the number of violations but

were in fewer cases the same amount regardless whether the violation was the first, second or third time.

Some had a mixture of non-monetary and monetary sanctions. Another common practice was to give a

monetary sanction the first two times and then dismissal the third time. There were also mixed penalties

such as combinations of monetary and additional work obligations or a monetary sanction combined with

last warning. There were also a lot of variation in the monetary amounts that had to be paid across groups

and the variation from first to second and third violation by the same person. It was the rule rather than the

exception to have bylaws with penalties for violations (80-93% of the groups), while 51-82% of all groups

had graduated sanctions for the four types of violations.

Ostrom’s DP 6: Conflict resolution system

About 25% of the youth groups have experienced at least one dispute and about 12% have experienced a

serious dispute. Disputes between the group and some outsiders were more likely to be of the serious type

than disputes within the group. 82.9% of the disputes within the group were resolved within the group itself

while only 19.2% of the disputes between the group and outsiders were resolved among the parties

themselves. As much as 39.7% of the disputes between the groups and outsiders were still unresolved at

the time of the interview while only 3.4% of the within-group disputes were still unresolved. There was a

high level of satisfaction (96.6% were satisfied) with how within-group disputes had been resolved while

23.4% were not satisfied with how the disputes between the group and some outsiders had been resolved.

We can conclude that local informal conflict resolution works well for within-group disputes. Because only

in few cases the groups had to get help from community (tabia) or district (woreda) officials to resolve

disputes.

Ostrom’s DP 7: Institutional recognition: Minimum rights to organize

The youth groups in Tigray have been formed through a formal institutional arrangement as primary

cooperatives and therefore fall under the law of cooperatives. The group is as such an official registered

unit of business with an exclusive number of members, that is provided a demarcated resource it is

responsible for managing in a sustainable way (except for mining where it has a time-limited right to extract

the non-renewable resource). The group has to have a business plan, have an officially elected leadership

team, can apply for joint loans, and to be audited by the cooperative organization. Still, the groups are given

the authority to organize themselves and have their own bylaws. The groups are monitored and provision

of loans and legal documents for their land entitlement depends on good performance and compliance with

the rules for resource management (taking good care of rehabilitated lands). We therefore conclude that the

groups get the minimum rights to organize themselves (Design Principle 7).

Ostrom’s DP 8: Nested enterprises: Common Pool Resource that is part of a larger system

The allocation of rehabilitated communal lands to youth groups in Tigray is clearly a part of a larger plan

for sustainable land management at community, district and regional levels. Parts of the areas allocated to

youth groups are “area exclosures”5 that the communities at an earlier stage agreed to protect from resource

extraction while they also invested in the conservation of the areas by building various forms of soil and

water conservation structures. Some areas have also been planted with trees. Other areas are rehabilitated

gulley areas. The rehabilitation has thus typically involved labor-intensive investments where labor has

been mobilized through compulsory labor provided by community members, through Food-for-work and

Cash-for-work activities under the Productive Safety Net Program (PSNP) or other food security and

sustainable land management programs.

We develop a more concise set of variables that should capture the variation in the extent to which the first

six DPs are implemented. These are summarized in Table 6. Table 6 below summarizes the final indicator

variables used to assess the DPs in the following regression models.

Table 6. Variables used as indicators for Ostrom’s Design Principles in regression models Variable Obs Mean Std. Dev. Min Max

DP1: Encroachment control system in place, dummy 742 0.941 0.236 0 1

DP2: Bylaw regulating sharing arrangements, dummy 742 0.973 0.162 0 1

DP3: Bylaw regulating frequency of meetings, dummy 742 0.949 0.221 0 1

DP3b: All members involved in decisions, dummy 742 0.708 0.455 0 1

DP4: Number of bylaws regulating management 741 3.497 1.044 0 4

DP5: Number of bylaws stipulating graduated sanctions 742 2.741 1.405 0 4

DP6: Satisfactory conflict resolution system, dummy 742 0.970 0.170 0 1

Source: Own census data.

The table demonstrates a high degree of compliance with the DPs among the youth groups. While there is

100% compliance with DP7 and DP8, the extent of deviation from 100% for the other DPs are used to

assess their possible effect on or correlation with the performance indicators. The DP variables in Table 6

may, however, also be closely correlated with each other and therefore create problems with

multicollinearity in the regressions. We therefore assess their correlation coefficients in Table A 24 in the

appendix. We also assessed the effects on significance levels from removing some of the most closely

correlated DPs but this did not make any big difference in the significance of the remaining DPs. The

correlation therefore appeared tolerable.

5 “Area exclosure” is a term used in Ethiopia and is an area that the community has agreed to protect from resource

extraction by humans and their livestock. The purpose has been to rehabilitate degraded communal lands. These areas

have also been called ‘enclosures’. The communities have typically established a guarding system to protect such

areas.

Next, we assess whether these DP indicator variables can be considered as “pre-determined” relative to the

outcome variables that we are assessing their correlation with. Assessing them one by one we find the

following. DP1 is assessed during the survey and we cannot rule out that the encroachment control system

to some extent has responded to actual exposure to violations. Such controls have a higher likelihood of

being imposed where encroachment is a problem and this may cause an underestimate of the effect of

imposing such controls. DP2, DP3, DP4 and DP5 are bylaws that were established by the youth groups

when they were established. We investigated whether these bylaws have been changes after their initial

establishment but there were hardly any such adjustments. We can therefore be confident that these

represent “pre-determined” variables. DP3b is on the other hand based on an assessment at the time of the

survey and indicates that decisions have not been delegated so much to the elected members or the leader.

This variable cannot therefore be considered to be pre-determined and the causality could go both ways.

DP6 is also an assessment at the time of the survey and may depend on the extent to which weaknesses in

the conflict resolution system have been revealed through exposure to conflicts. We therefore have to be

careful with the interpretation of correlations for this variable.

5. Estimation strategy

The key performance indicators are outlined in Table 3 and Table 4. For the first of the indicators, exposure

to conflicts we use two types of models; an ordered probit model and a multinomial logit model. This allows

us to test whether exposure to less serious disputes is related to different variables than exposure to more

serious disputes. For the other performance indicators, except the share of the initial group members that

still are in the group, we use ordered probit models. A fractional response model is used for the member

share staying models. The ordered probit models include the ranked degree of effect of poor cooperation

on performance (group leaders’ assessment), the level of trust (group leaders’ assessment), group leaders’

ranked performance assessment by the groups, and the youth association’s assessment of group

performance. These performance indicators are assumed to be functions of the DPs and possibly a number

of other control variables. We first run parsimonious models without the control variables. The control

variables are shown in Table 4. Some of them are potentially endogenous and require careful treatment.

Finding valid and strong instruments to predict them is difficult.

Finally, we assess group performance using group income per member as a dependent variable. In these

models we have added potential endogenous variables in a stepwise fashion to check the robustness of the

results for the DPs. The basic hypothesis is that the DPs are enhancing the performance of the groups and

that a significant correlation with the performance indicators may imply a causal positive effect on

performance for the pre-determined DPs. For the potentially more endogenous DPs we discuss the results

more cautiously in terms of there possibly being reverse causality or some other form of interaction.

An assessment of the potential endogeneity of the DPs was implemented by regressing the DPs on the same

set of control variables. The results are presented in Appendix Table A 25. It shows that certain DPs were

more likely to have been implemented in certain woredas. It also shows a higher R2 for the DPs that were

more likely to have changed after the formation of the groups (DP1, DP3b, DP6). We have resorted to

careful interpretation of the results when potentially endogenous variables are included, whether these are

the DPs or the control variables.

The DPs are also potentially correlated and we have assessed the degree of such correlation. It is substantial

between some of the DPs and we ran specifications where some of the highly correlated DPs were dropped

to assess how and whether this affected significance levels and the coefficients of the related included DPs.

Such multicollinearity did not affect the robustness of our key results and conclusions

6. Results and discussion

The results for the parsimonious DP and performance models without other control variables are presented

in Table 7. Positive significant coefficients for DPs in the member stay and the income per member models

can indicate that DPs enhance performance. Negative significant coefficients for DPs in the other models

indicate that DPs enhance performance in the wanted direction. DP1 (encroachment control system

established) is significantly associated with better performance for five out of six performance indicators.

DP2 (Bylaw regulating sharing arrangements) and DP4 (Number of bylaws regulating management) are

not significantly associated with any of the performance indicators. DP3 (Bylaw regulating frequency of

meetings) is significantly related with, possibly enhancing, two of the performance indicators (the rating

by youth group leaders and Youth Associations). DP3b (All members involved in decisions) is significantly

related with three indicators but only two of these are in the enhancing direction (indicators for poor

performance and trust). DP5 (number of graduated sanctions) is significantly related with four of the

indicators, three of which are in the expected direction. It seems to have contributed positively to (or is

positively correlated with) higher share of members staying on in the groups but is negatively correlated

with income per member. A higher number of members staying also implies that more members have to

share the income from the group activity and this may partly explain this finding. Overall, for seven DP

measures and six performance indicators we found 14 significant positive relationships enhancing

performance and three that were significant and negative. The high compliance with the DPs resulted in

few observations where the DPs were unfulfilled. This may have limited the number of significant positive

effects such as for DP2 where compliance was 97.3%.

14

Table 7. Parsimonious models for DPs versus performance indicators Member

Stay share

Poor coop

affects

performance

Trust Own rating Youth

Association

Rating

Group

income per

member

DP1: Encroachment control system in place 0.328c -0.303 -0.703a -0.752a -0.766b 5.013a

DP2: Bylaw regulating sharing arrangements 0.415 0.330 -0.186 -0.080 -0.122 -0.480

DP3: Bylaw regulating frequency of

meetings

0.133 0.031 0.236 -0.449d -0.545c 1.946

DP3b: All members involved in decisions -0.072 -1.020a -0.443a 0.004 0.326a 0.398

DP4: Number of bylaws regulating

management

-0.080 -0.011 -0.095 0.046 0.029 0.334

DP5: Number of graduated sanctions 0.109b -0.096c -0.024 -0.001 -0.075c -0.425c

DP6: Satisfactory conflict resolution system 0.162 0.993a -0.651c -0.427 -0.699c 1.477

Constant 0.232

-5.884c

Cut1 constant

-0.930c -1.848a -1.861a -3.172a

Cut2 constant

-0.500 -0.176 -0.829 -1.853a

Cut3 constant

0.190 -0.011 -0.389

Cut4 constant

0.861d 0.624 0.230

Sigma constant

5.629a

Wald chi2 19.639 89.559 54.093 23.213 49.572

Log likelihood

-1554.391

Prob > chi2 0.006 0.000 0.000 0.002 0.000 0.001

Number of observations 741 736 740 740 739 733 Source: Own census data. Significance levels: d < 0.10, c < 0.05, b < 0.01, a < 0.001. Fractional probit model with robust standard errors is used for the member stay model.

Censored tobit with robust standard erros is used in the income per member model. Ordered probit models with robust standard errors are used in the remaining models.

15

Table 8 presents the results from the performance-related models where additional controls are included

as a robustness check. The responses for the dependent variables, except the member share of the initial

group members staying in the group, are shown in Table 3. The distribution of the member share of the

initial group staying is shown in Figure 1. The first model in

Table 8 assesses factors associated with whether poor cooperation in the group affects its performance.

DP3b is negatively associated with group cooperation (significant at 0.1% level). More active participation

by all members in group decisions therefore seems to be associated with more cooperation problems in the

groups and these negatively affect the performance of the group. However, here could be reverse causality;

if performance is affected group members may engage themselves more in solving the problem. DP5

(number of graduated sanctions) is also negatively associated with (significant at 1% level) performance of

the groups. It is possible that tougher sanctions create more frustrations and more problems in the groups.

DP6 (satisfactory conflict resolution system in place) is positively associated with group performance

(significant at 5% level). The dynamics in the groups related to the DPs appear to be ambiguous.

A look at the other controls provide some additional insights. Exposure to conflicts severely affects, or is

correlated with, group cooperation and performance (significant at 1% level for both less serious and serious

disputes). Input access problems are strongly affecting group cooperation and performance (significant at

0.1% level). Early established groups are more likely to have cooperation and performance problems

(significant at 1% level).

We next assess the rating of the level of trust among group members (higher trust is associated with a lower

number). DP1 (Encroachment control system in place) is associated a higher level of trust within the group

(significant at 0.1% level). DP3b (active participation by all group members in decision) is also significantly

(at 0.1% level) associated with a higher level of trust in the groups. None of the other DPs were significant.

Exposure to less serious and serious conflicts are associated with significantly lower (at 0.1% levels) within

group trust. Groups that were allowed to form themselves (self-selection of members) demonstrate

significantly (at 0.1% level) higher level of internal trust. A change in key group positions in the groups is

associated with a significantly (at 1%) lower level of trust. The level of trust is also significantly lower in

two of the districts while there are no significant differences between types of activities.

The next model uses the group leaders’ rating of the performance of their group relative to other groups on

a scale from one to five, with one being the best (see Table 3 for the distribution). DP1 (Encroachment

control system established) is significantly (at 0.1% level) related to better ranking of group performance.

None of the other DPs are significant in this model. Exposure to less serious or serious conflicts are

significantly (at 0.1% levels) associated with poorer performance rating of the groups.

Irrigation/horticulture groups had a significantly (at 1% level) better rating than other types of groups. More

recently established groups also received a significantly (at 5% level) better rating while one of the districts

had a better rating of the groups there.

We next compare the Youth Association’s rating of the groups (a lower number indicating a higher/better

rating). It is possible that youth group leaders overstate the performance of their groups. The distribution

of the responses in Table 3 points in this direction. On the other hand, group leaders may also be better

informed about internal conditions in the groups than the Youth Association, which has representatives

located in the tabia who are responsible for following up the groups.

Table 8 demonstrates that DP1 (encroachment control system in place) is highly significant (at 0.1% level)

and positively associated with the group performance rating. DP3 (having bylaw regulating frequency of

meetings) is also significantly (at 5% level) associated with a better rating while DP3b (all members

involved in group decisions) is significantly negatively associated with the performance rating in contrast

to DP3. The latter variable is more endogenous and the finding may relate to what is observed related to

DP3b in the previous models on conflicts and cooperation versus performance. DP5 (number of graduated

sanctions) is significantly (at 0.1 level) associated with group ranking and the same is the case for DP6

(satisfactory conflict resolution system in place) which is significant at 5% level. This gives a good

correspondence between the DPs and group performance as judged by the Youth Association.

Contrary to the other models, the less serious and serious conflict variables are not significantly related to

the rating by the Youth Association. Significant differences were found between some of the districts and

like the rating by group leaders irrigation /horticulture groups were rating as performing significantly (at

5% level) better than other groups. Forest groups were rated to perform poorer than other groups (significant

at 10% level only). Change in key group positions and larger area per group members were also associated

with higher ranking (both of these variables were significant at 10% levels only).

Table 8. Ordered probit models for group cooperation, group trust, own performance rating and

rating by the Youth Association Poor

cooperation

affects

performance

Trust Own

rating

Youth

Association

Rating

DP1: Encroachment control system in place -0.163 -0.812a -0.772a -0.805a

DP2: Bylaw regulating sharing arrangements 0.375 -0.433 -0.072 -0.346

DP3: Bylaw regulating frequency of meetings 0.405 -0.035 -0.466d -0.564c

DP3b: All members involved in decisions -1.252a -0.386a -0.022 0.284b

DP4: Number of bylaws regulating management -0.063 -0.005 0.068 0.090

DP5: Number of graduated sanctions -0.122c -0.068 -0.024 -0.121b

DP6: Satisfactory conflict resolution system 0.770c -0.404 0.119 -0.542c

Establishment year & month 0.104b 0.053 -0.063c -0.013

Initial member number -0.008c -0.004 -0.003 -0.001

Female share in group at start 0.314 -0.347 -0.339 0.219

Gender of group leader, 1=Female 0.485b 0.013 0.013 -0.153

Distance to road, km -0.024 0.007 -0.015 -0.002

Distance to market, km -0.015 0.000 0.011 0.011

Distance to home, min. km 0.055 -0.002 0.019 0.038

Input access problems, dummy -0.511a -0.153 0.117 0.042

Baseline: No conflicts 0.000 0.000 0.000 0.000

Serious conflict in group, dummy -0.499b 0.472b 0.538a 0.111

Less serious conflict in group, dummy -0.434b 0.615a 0.444a -0.014

Area allocated per person in group 0.491d -0.185 -0.390d -0.439d

Number of returning migrants 0.027 -0.024 -0.013 -0.013

Change in key group positions, dummy -0.182 0.261c -0.158 -0.157

Self-selection of group members, dummy 0.249 -0.535a -0.075 0.274c

District dummies: Base=Raya Azebo 0.000 0.000 0.000 0.000

Degua Tembien 0.159 -0.103 -0.178 0.225

Seharti Samre 0.335d 0.480b -0.208 0.539a

Kilite Awlalo -0.062 0.053 -0.475a -0.345c

Adwa 0.501b 0.293d -0.096 0.398b

Activity dummies: Base=Animal rearing 0.000 0.000 0.000 0.000

Beekeeping 0.059 -0.122 0.103 0.149

Forest -0.198 -0.246 0.025 0.350d

Irrigation/Horticulture 0.218 -0.063 -0.527a -0.304c

Mining 0.132 -0.213 -0.197 -0.099

Other 0.941 -1.447d -0.905 -0.005

Cut 1 Constant 207.546b 104.046 -128.856c -28.367

Cut 2 Constant 208.034b 105.863 -127.760c -26.939

Cut 3 Constant 106.225 -126.907c -25.398

Cut 4 Constant 106.896 -126.257c -24.765

Wald chi2 194.619 134.823 107.778 152.411

Prob > chi2 0.000 0.000 0.000 0.000

Number of obs. 722 725 726 725 Source: Own census data. Dependent variable categorizations: Group cooperation problem: 1=Yes, very much, 2=To some

extent, 3=No cooperation problem, Trust: 1=Very high, 2=Quite high, 3=ok, 4=Not so good, 5=Very poor, Own performance

rating and Youth Association rating: 1=Very good, 2=Good, 3=Average, 4=Below average, 5=Poor performance. Significance

levels: d < 0.10, c < 0.05, b < 0.01, a < 0.001.

21

Table 9 assesses whether the DPs affect share of remaining initial members in the group and the income

that youth group members obtain from their youth group activity. The dependent variable in the income

models is the log of the income per current youth group member in 2015 in Ethiopian Birr (EB). Descriptive

statistics of the variation in income per current youth group member across the main group activities is

shown in Table 4. Forest and beekeeping groups have much lower incomes than the other groups. It takes

long time before planted trees can be harvested. 2015 was a drought year and this may also have affected

negatively the production of honey by beekeeping groups.

We have included the member stay share model with the income models to facilitate a better joint inspection

of these models. A stepwise approach is used to assess the importance of the following potential endogenous

variables. These are occurrence of more or less serious conflicts, number of dropped out members (which

relates to the member stay share as one of the performance indicators), change in elected board of group

members (dummy=1 if such a change has taken place), and gender of group leader (dummy=1 for female

leader). Finding good instruments to predict these is difficult. We therefore resort to running models with

and without these and careful interpretation of the results.

Table 9 shows a very strong and highly significant (at 0.1% level) relationship between DP1 (encroachment

control system in place). The direction of causality could go both ways in this case. With a more valuable

production activity which generates more income for the group there may also be higher risk of

encroachment and theft and making guarding of the area important. Among the other DPs only DP5

(number of graduated sanctions) was significant (at 5 and 10% levels) and it had a negative sign. A higher

number of graduated sanctions was associated with lower income per group member. This could indicate a

dis-incentive effect from graduated sanctions. The signs of all the other DPs were positive but insignificant.

Among the other variables, we see that establishment year and month is highly significant and negative in

the income models and positive in the member stay share models. This makes sense as it takes time before

the youth group production activities start to generate income. Dropout rates may also increase over time

and there may have been more organizational problems in relation to the establishment of the earliest groups

due to less experienced local administrations. Lower income is particularly associated with the forest groups

for tree production takes longer time than other activities to generate income. Beekeeping also gave

significantly lower income than other activities except forestry. Input access problems (dummy) was

significantly negatively associated with income and 66% of the groups stated to have such a problem. Lower

income is not directly related to distance to road or distance to market as these variables are insignificant.

Income is positively related to the area allocated per initial group member (significant at 10 and 5% levels).

Income per current member is also highly significantly (at 0.1% level) related to the reduction in number

of group members since start. A higher dropout of members gives more income per remaining members.

Here could also be a selection effect with more hard-working members remaining. The initial female share

in the group is negatively associated with income per initial member. This could be because some of the

investment activities are physically demanding and more difficult for females to contribute to. Some groups

allowed females to contribute less to such activities without changing the principle of equal sharing of

income. Finally, we see, somewhat surprisingly, that groups that have less serious disputes have

significantly (at 0.1% level) higher income per member while groups with serious disputes also have higher

income (significant at 10% only) than groups with no disputes. This could be because higher income creates

more competition and disagreements within groups (reverse causality). Another finding, that groups that

have had a change in key positions since start (27% of the groups have had such a change), had significantly

higher income per member. Here also causality could go both ways. Higher income could cause more

internal controversies and a demand for change of persons in key positions. Such a change could also lead

to better group management and higher income.

23

Table 9. Fractional probit model for member stay share and Tobit models for log of individual

income per current member in 2015 from group activity Member stay

share

Logindinc1 Logindinc2 Logindinc3

DP1: Encroachment control system in place 0.296 5.548a 5.566a 4.999a

DP2: Bylaw regulating sharing arrangements 0.217 -1.781 -1.791 -1.913

DP3: Bylaw regulating frequency of meetings 0.189 0.931 1.150 0.810

DP3b: All members involved in decisions -0.013 0.459 0.285 0.152

DP4: Number of bylaws regulating

management

-0.047 0.364 0.262 0.350

DP5: Number of graduated sanctions 0.086c -0.385c -0.315d -0.295d

DP6: Satisfactory conflict resolution system 0.102 0.934 1.054 1.457

Establishment year & month 0.097a -0.893a -0.836a -0.655a

Initial member number 0.000 -0.004 -0.019 -0.023

Female share at start of group 0.032 -2.330c -2.484c -2.792c

Distance to road, km -0.156 -0.013 -0.017 0.021

Distance to market, km 0.018 -0.013 -0.012 -0.031

Distance to home, min. km 0.012d -0.287d -0.260 -0.262

Input access problems, dummy -0.027 -1.171c -1.122c -1.339b

Area allocated per person in group 0.057 1.919d 2.593c 1.985d

Number of returning migrants 0.008 0.042 0.063 0.055

District dummies: Base=Raya Azebo

Degua Tembien 0.072 2.287b 2.070c 1.642c

Seharti Samre 0.501a 0.594 0.823 0.841

Kilite Awlalo -0.058 1.245d 1.327d 1.725c

Adwa 0.582a 0.278 0.275 0.139

Activity dummies: Base=Animal rearing

Beekeeping -0.032 -2.543a -2.501a -2.543a

Forest -0.017 -5.220a -4.986a -4.834a

Irrigation/Horticulture 0.116 -0.893 -0.525 -0.388

Mining -0.130 0.035 -0.037 -0.015

Other -0.200 0.422 0.206 0.174

Number of dropped out members

0.147a 0.123a

Gender of group leader, 1=Female

0.572 0.828

Self-selection of group members -0.143

Baseline: No conflicts

Serious conflict in group, dummy -0.316c

1.039d

Less serious conflict in group, dummy -0.108

2.025a

Change in key group positions, dummy -0.356a

2.187a

Constant -195.891a 1797.448a 1681.809a 1317.806a

Sigma constant 5.197a 5.141a 4.986a

Log likelihood -1490.106 -1477.733 -1458.721

Wald chi2 114.140

Prob. > chi2 0.000 0.000 0.000 0.000

Number of obs. 726 721 719 719

Number of left censored obs. 315 315 315 Source: Own census data. Dependent variable: Log (Income per current group member (EB) in 2015+1). The table shows

marginal effects. Significance levels: d < 0.10, c < 0.05, b < 0.01, a < 0.001.

24

7. Conclusion

Ostrom (2010) warns against being too optimistic about being able to solve social dilemmas as there are

many examples of failures as well, and acknowledges that further research is needed. We contribute to this

research in terms of assessing the success of kick-started youth groups in terms of their ability to cooperate

and self-manage themselves as custodians of an allocated land resource that is intended to be a source of

livelihood and joint business. Our study revealed a high degree of compliance with Ostrom’s Design

Principles across the youth groups. When relating the Design Principles to a number of performance

indicators, we found that DP1 (having an encroachment control system in place) was the most significant

among the DPs. This is in terms of being associated with better performance (significant for five of six

indicators in the parsimonious models without controls and significant in four of six in the models with an

extended set of controls). Overall, for the seven DP measures and six performance indicators we found 14

significant effects or correlations in line with DPs enhancing performance in the parsimonious models while

two of those became insignificant when we added the set of controls. There were three significant

relationships between the DPs and the performance indicators that had the opposite sign. Five of the seven

DP measures were significantly correlated with the Youth Association’s ranking of the performance of the

groups and one of these had the “wrong” sign (DP3b - All members involved in decisions), but only DP1

was significantly positively correlated with income of group members from their joint production activity.

Self-selection of group members, used by 80% of the groups, is associated with higher trust among

members. Some results seemed surprising. For example, exposure to disputes and change in key group

positions were associated with lower trust but also with higher incomes from the groups’ joint production

activities. This latter result may partly be explained by larger dropout rates for groups with such exposure

to serious disputes and changes in key group positions as a lower number of group members lead to higher

income per remaining member. However, how much this can explain of these income effects needs further

inspection.

Overall, we think the youth group model we have studied shows promise as an approach to engaging

landless and unemployed youth in productive activities and as environmental custodians. Giving youth

groups self-organization responsibilities, opportunities, as well as obligations as a primary cooperative and

environmental caretaker unit is an approach that we recommend for testing also elsewhere, given the

growing problem of youth unemployment and landlessness. This can also contribute to political stability

and keeping youth out of trouble and making them into the valuable resource for society they deserve to

be.

25

References

Agrawal, A. (2001). Common property institutions and sustainable governance of resources. World

Development 29(10), 1649-1672.

Baland, J. M., & Platteau, J. P. (1996). Halting degradation of natural resources: Is there a role for rural

communities? Food & Agriculture Organization, Rome.

Cox, M., Arnold, G. and Tomas, S.V. (2009). Design principles are not blue prints, but are they robust? A

meta-analysis of 112 studies. Lincoln Institute of Land Policy Working Paper.

Ostrom, E. (1990). Governing the commons. Cambridge University Press.

Ostrom, E. (2010). Beyond markets and states: polycentric governance of complex economic systems.

American Economic Review 100(3), 641–672.

Ostrom, E. (2010). Polycentric systems for coping with collective action and global environmental change.

Global Environmental Change 20(4), 550-557.

Ostrom, E. (2010). Analyzing collective action. Agricultural Economics 41, 155-166.

Ostrom, E., Walker, J., & Gardner, R. (1992). Covenants with and without a Sword: Self-governance Is

Possible. American Political Science Review 86(2), 404-417.

Ostrom, E. and Basurto, X. (2011). Crafting analytical tools to study institutional change. Journal of

Institutional Economics 7(3), 317-343.

Poteete, A. R., and Ostrom, E. (2008). Fifteen years of empirical research on collective action in natural

resource management: struggling to build large-N databases based on qualitative research. World

Development 36(1), 176-195.

Wade, R. (1988). The management of irrigation systems: How to evoke trust and avoid prisoner's dilemma.

World Development 16(4), 489-500.

26

APPENDICES

Appendix 1A: Details on adherence to Ostrom’s DPs

Ostrom’s DP 1: Clearly defined borders

In Table A 1 we assess whether the land area that has been allocated to youth groups is clearly demarcated.

The groups are grouped by the main activity they have on their land to assess whether there are systematic

differences between these.

Table A 1. Assessment of Ostrom’s Design Principle 1: Clear border demarcation, by type of

activity Animal

rearing

Bee

keeping

Forest Irrigation/

Horticulture

Mining Other Total

Very clearly

demarcated and fenced

Obs. 40.0 23.0 4.0 38.0 2.0 2.0 109.0

% 29.2 14.5 9.5 39.6 0.7 20.0 14.8

Clearly demarcated

but not fenced

Obs. 70.0 110.0 29.0 46.0 206.0 0.0 461.0

% 51.1 69.2 69.1 47.9 69.8 0.0 62.4

Partly well demarcated Obs. 10.0 10.0 4.0 2.0 2.0 0.0 28.0

% 7.3 6.3 9.5 2.1 0.7 0.0 3.8

No clear borders for

part of the area

Obs. 11.0 16.0 5.0 10.0 85.0 1.0 128

% 8.0 10.1 11.9 10.4.0 28.8 10.0 17.3

Does not apply Obs. 6.0 0.0 0.0 0.0 0.0 7.0 13.0

% 4.4 0.0 0.0 0.0 0.0 70.0 1.8

Total Obs. 137.0 159.0 42.0 96.0 295.0 10.0 739.0

% 100.0 100.0 1000.0 100.0 100.0 100.0 100.0

Source: Own census data. Test for significant difference between activities: Pearson chi2(20) = 454.14, Pr < 0.0001.

27

Table A 2. Is there traffic by outsiders going through the area allocated to the youth group? By

district Raya

Azebo

Degua

Tembien

Seharti

Samre

Kilite

Awlalo

Adwa Total

Yes, a path/road goes through Obs. 57.0 3.0 11.0 43.0 10.0 124.0

% 28.8 4.1 10.2 26.7 5.1 16.8

It is common by outsiders

to walk through the area

Obs. 48.0 6.0 18.0 69.0 11.0 152.0

% 24.2 8.2 16.7 42.9 5.6 20.6

Livestock of outsiders

commonly enter the area

Obs. 18.0 5.0 12.0 13.0 29.0 77.0

% 9.1 6.9 11.1 8.1 14.7 10.5

Uncommon, but it happens Obs. 30.0 13.0 18.0 28.0 57.0 146.0

% 15.2 17.8 16.7 17.4 28.9 19.8

No, it is well protected and

no traffic by outsiders

Obs. 42.0 46.0 41.0 7.0 89.0 225.0

% 21.2 63.0 38.0 4.4 45.2 30.5

Does not apply Obs. 3.0 0.0 8.0 1.0 1.0 13.0

% 1.5 0.0 7.4 0.6 0.5 1.8

Total Obs. 198.0 73.0 108.0 161.0 197.0 737.0

% 100.0 100.0 100.0 100.0 100.0 100.0

Source: Own census data. Test for significant difference between districts: Pearson chi2(20) = 243.76, Pr < 0.0001.

Table A 3. Is there traffic by outsiders going through the area allocated to the youth group? By

main activity type Animal

rearing

Beekeeping Forest Irrigation/

Horticulture

Mining Other Total

Yes, a path/road goes

through

Obs. 6.0 5.0 1.0 4.0 107.0 1.0 124.0

% 4.4 3.2 2.4 4.2 36.4 10.0 16.8

It is common by outsiders

to walk through the area

Obs. 14.0 8.0 2.0 5.0 123.0 0.0 152.0

% 10.2 5.1 4.8 5.2 41.8 0.0 20.6

Livestock of outsiders

commonly enter the area

Obs. 14.0 23.0 7.0 7.0 26.0 0.0 77.0

% 10.2 14.6 16.7 7.3 8.8 0.0 10.5

Uncommon, but it

happens

Obs. 31.0 44.0 18.0 21.0 32.0 0.0 146.0

% 22.6 27.9 42.9 21.9 10.9 0.0 19.8

No, it is well protected

and

no traffic by outsiders

Obs. 66.0 78.0 14.0 59.0 6.0 2.0 225.0

% 48.2 49.4 33.3 61.5 2.0 20.0 30.5

Does not apply Obs. 6.0 0.0 0.0 0.0 0.0 7.0 13.0

% 4.4 0.0 0.0 0.0 0.0 70.0 1.8

Total Obs. 137.0 158.0 42.0 96.0 294.0 10.0 737.0

% 100.0 100.0 100.0 100.0 100.0 100.0 100.0

Source: Own census data. Test for significant difference between activities. Pearson chi2(25) = 678.33, Pr < 0.0001.

28

Table A 4. Is the group experiencing any illegal harvesting by outsiders in the allocated land

area? Raya

Azebo

Degua

Tembien

Seharti

Samre

Kilite

Awlalo

Adwa Total

It is frequent (>1 per week) Obs. 6.0 1.0 1.0 0.0 0.0 8.0

% 3.1 1.4 0.9 0.0 0.0 1.1

It happens now and then

(>1 per month)

Obs. 11.0 0.0 6.0 4.0 12.0 33.0

% 5.6 0.0 5.6 2.5 6.1 4.5

It happens rarely (<1 per

month)

Obs. 8.0 5.0 4.0 1.0 26.0 44.0

% 4.1 6.9 3.7 0.6 13.1 6.0

It happens very rarely

(<1 per year)

Obs. 21.0 16.0 9.0 14.0 49.0 109.0

% 10.7 21.9 8.3 8.8 24.8 14.8

Has never happened since start

of the group/area was allocated

Obs. 150.0 51.0 88.0 141.0 111.0 541.0

% 76.5 69.9 81.5 88.1 56.1 73.6

Total Obs. 196.0 73.0 108.0 160.0 198.0 735.0

% 100.0 100.0 100.0 100.0 100.0 100.0

Source: Own census data. Test for significant difference between districts: Pearson chi2(16) = 83.42, Pr < 0.0001.

Table A 5. Frequency of illegal harvesting vs type of activity Animal

rearing

Bee-

keeping

Forest Irrigation/

Horticulture

Mining Other Total

It is frequent (>1 per

week)

Obs. 3.0 1.0 0.0 1.0 3.0 0.0 8.0

% 2.2 0.6 0.0 1.0 1.0 0.0 1.1

It happens now and then

(>1 per month)

Obs. 6.0 5.0 3.0 6.0 13.0 0.0 33.0

% 4.4 3.1 7.1 6.3 4.4 0.0 4.5

It happens rarely

(<1 per month)

Obs. 13.0 15.0 8.0 6.0 1.0 1.0 44.0

% 9.6 9.4 19.1 6.3 0.3 12.5 6.0

It happens very rarely

(<1 per year)

Obs. 24.0 36.0 10.0 17.0 22.0 0.0 109.0

% 17.7 22.6 23.8 17.7 7.5 0.0 14.8

Has never happened since

start of the group/area was

allocated

Obs. 90.0 102.0 21.0 66.0 255.0 7.0 541.0

% 66.2 64.2 50.0 68.8 86.7 87.5 73.6

Total Obs. 136.0 159.0 42.0 96.0 294.0 8.0 735.0

% 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Source: Own census data. Test for significant difference between activities: Pearson chi2(20) = 74.52, Pr < 0.0001.

We see from Tables A4 and A5 that less than 6% of the youth groups experience illegal harvesting on their

land more frequently than monthly and only 1.1% experience it more frequently than weekly. This shows

that illegal harvesting is not a big problem. Table 13 shows the extent of guarding that is used by the youth

groups to protect their area. Most youth groups have a guarding system and we can conclude that the

resource base of the youth is well protected from extraction by outsiders.

29

Table A 6. Frequency of illegal harvesting versus guarding of the land against outsiders (related

to Ostrom’s DP 1: Clear demarcation against outsider intrusion Frequency of illegal harvesting by

outsiders in the allocated area

Continuously

guarding by

rotating

responsibility

Guarding

during

daytime

Hired a

guard to

protect the

area

No guard is

considered

necessary

Total

It is frequent (>1 per week) Obs. 6.0 1.0 0.0 1.0 8.0

% 1.7 1.4 0.0 0.6 1.1

It happens now and then

(>1 per month)

Obs. 22.0 3.0 3.0 5.0 33.0

% 6.2 4.2 2.0 3.2 4.5

It happens rarely (<1 per month) Obs. 15.0 6.0 19.0 4.0 44.0

% 4.2 8.3 12.8 2.6 6.0

It happens very rarely

(<1 per year)

Obs. 56.0 16.0 33.0 4.0 109.0

% 15.7 22.2 22.2 2.6 14.8

Has never happened since start

of the group/area was allocated

Obs. 258.0 46.0 94.0 143.0 541

% 72.3 63.9 63.1 91.1 73.6

Total Obs. 357.0 72.0 149.0 157.0 735.0

% 100.0 100.0 100.0 100.0 100.0

Source: Own census data. Test for significant difference between districts: Pearson chi2(12) = 59.00, Pr < 0.0001.

Table A 7. Is the current system for controlling encroachment functioning satisfactorily?

(Ostrom’s DP 1/5?) Raya

Azebo

Degua

Tembien

Seharti

Samre

Kilite

Awlalo

Adwa Total

No Obs. 25.0 5.0 9.0 2.0 3.0 44.0

% 12.6 6.9 8.3 1.2 1.5 6.0

Yes Obs. 170.0 68.0 99.0 158.0 194.0 689.0

% 85.9 93.1 90.8 98.1 98.0 93.2

Some modification is needed Obs. 3.0 0.0 1.0 1.0 1.0 6.0

% 1.52 0.0 0.92 0.6 0.5 0.8

Total Obs. 198.0 73.0 109.0 161.0 198.0 739.0

% 100.0 100.0 100.0 100.0 100.0 100.0

Source: Own census data. Test for significant difference between districts: Pearson chi2(8) = 32.66, Pr < 0.0001.

From this we conclude that Ostrom’s DP 1 (Clearly demarcated and protected borders for the resource

and clearly defined group with rights) is satisfied for 93% of the groups.

Ostrom’s DP 2: Sharing of benefits and costs/congruence between appropriation rules and

provision rules

Table A 8 shows that 97.6% of the groups have a bylaw that regulates the sharing of responsibilities and

incomes from the group activities.

30

Table A 8. Is the sharing of work and responsibilities in the group and the sharing of income

regulated by the bylaw of the group? Raya

Azebo

Degua

Tembien

Seharti

Samre

Kilite

Awlalo

Adwa Total

No Obs. 14.0 0.0 0.0 3.0 1.0 18.0

% 7.0 0.0 0.0 1.9 0.5 2.4

Yes Obs. 185.0 73.0 109.0 158.0 197.0 722.0

% 93.0 100.0 100.0 98.1 99.5 97.6

Total Obs. 199.0 73.0 109.0 161.0 198.0 740.0

% 100.0 100.0 100.0 100.0 100.0 100.0

Source: Own census data.

Table A 9. How is the work required on the allocated land shared among the youth group

members? (Ostrom’s DP 2) Raya

Azebo

Degua

Tembien

Seharti

Samre

Kilite

Awlalo

Adwa Total

Equal sharing for all Obs. 108.0 60.0 86.0 106.0 151.0 511.0

% 54.0 82.2 78.9 65.8 76.3 69.0

Different requirement for males

and females

Obs. 85.0 13.0 17.0 54.0 47.0 216.0

% 42.5 17.8 15.6 33.5 23.7 29.2

Sharing modified to ability of

each member

Obs. 3.0 0.0 0.0 1.0 0.0 4.0

% 1.5 0.0 0.0 0.6 0.0 0.5

Sharing based on the individual

motivation & other activities

Obs. 3.0 0.0 6.0 0.0 0.0 9.0

% 1.5 0.0 5.5 0.0 0.0 1.2

Other, specify Obs. 1.0 0.0 0.0 0.0 0.0 1.0

% 0.5 0.0 0.0 0.0 0.0 0.1

Total Obs. 200.0 73.0 109.0 161.0 198.0 741.0

% 100.0 100.0 100.0 100.0 100.0 100.0

Source: Own census data. Test for significant difference between districts: Pearson chi2(16) = 67.13, Pr < 0.0001.

Table A 10. How is income from the group activities shared among group members? (Ostrom’s

DP 2). Raya

Azebo

Degua

Tembien

Seharti

Samre

Kilite

Awlalo

Adwa Total

Equally shared by all Obs. 198.0 73.0 102.0 159.0 198.0 730.0

% 99.0 100.0 93.6 99.4 100.0 98.7

Shares depend on effort

and gender

Obs. 2.0 0.0 7.0 1.0 0.0 9.0

% 0.5 0.0 6.4 0.6 0.0 1.2

Total Obs. 200.0 73.0 109.0 160.0 198.0 740.0

% 100.0 100.0 100.0 100.0 100.0 100.0

Source: Own census data. Test for significant difference between districts: Pearson chi2(8) = 31.93, Pr < 0.0001.

31

We see from Tables A9 and A10 that equal sharing of income is the rule for 98.7% of the youth groups. In

Table 18 we see that the responsibilities are shared equally in 56.7% of the groups only as female members

are allowed to do less of the heavy body work than males but they still get an equal share of the benefits.

Table A 11. Is there a gender difference in the sharing of responsibilities and benefits in the

group? (Ostrom’s DP 2). Raya

Azebo

Degua

Tembien

Seharti

Samre

Kilite

Awlalo

Adwa Total

No, all are equally treated Obs. 114.0 34.0 61.0 61.0 117.0 387.0

% 58.5 46.6 69.3 45.9 60.3 56.7

Females do less of the heavy

work but get equal benefit

Obs. 79.0 39.0 24.0 72.0 77.0 291.0

% 40.5 53.4 27.3 54.1 39.7 42.6

Females do less of the heavy

work and get less of the benefit

Obs. 0.0 0.0 3.0 0.0 0.0 3.0

% 0.0 0.0 3.4 0.0 0.0 0.4

Other, specify Obs. 2.0 0.0 0.0 0.0 0.0 2.0

% 1.0 0.0 0.0 0.0 0.0 0.3

Total Obs. 195.0 73.0 88.0 133.0 194.0 683.0

% 100.0 100.0 100.0 100.0 100.0 100.0

Source: Own census data. Test for significant difference between districts: Pearson chi2(12) = 43.99, Pr < 0.0001.

If we accept that females should be allowed to do less of the heavy work, we can state that 99% of the

groups satisfy DP 2.

Ostrom’s DP 3: Joint decision-making system.

95.8% of the youth groups stated to have their own written bylaw. Their bylaws also stated the frequency

of meetings in these groups that have a bylaw, see Table A 12.

Table A 12. Groups with bylaw that specified the frequency of meetings, by district (Ostrom’s

DP 3) Raya

Azebo

Degua

Tembien

Seharti

Samre

Kilite

Awlalo

Adwa Total

Weekly Obs. 90.0 23.0 31.0 79.0 13.0 236.0

% 50.0 31.5 31.3 50.3 6.8 33.7

Biweekly Obs. 52.0 22.0 28.0 52.0 85.0 239.0

% 28.9 30.1 28.3 33.1 44.3 34.1

Monthly Obs. 37.0 28.0 40.0 26.0 90.0 221.0

% 20.6 38.4 40.4 16.6 46.9 31.5

Other, explain Obs. 1.0 0.0 0.0 0.0 4.0 5.0

% 0.6 0.0 0.0 0.0 2.1 0.7

Total Obs. 180.0 73.0 99.0 157.0 192.0 701.0

% 100.0 100.0 100.0 100.0 100.0 100.0

Source: Own census data. Test for significant difference between districts: Pearson chi2(12) =120.73, Pr < 0.0001.

32

92.2% of the groups have penalties for non-participation in group meetings. There are actually graduated

sanctions against non-participation as well as late arrival to meetings. And these have been enforced.

Monetary sanctions are most common (86% of the sanctions). 43.6% of the groups had members that had

been penalized for absence from meetings and 36.9% of the groups had members that had been penalized

for late arrival to meetings. It is very clear that group membership implies strong participation obligations.

Close to 68% of the groups have meetings at least biweekly.

Table A 13. Who are the most dominant in making decisions in the group and enforcing

consensus decisions? (Ostrom’s DP 3) Raya

Azebo

Degua

Tembien

Seharti

Samre

Kilite

Awlalo

Adwa Total

Male group Obs. 8.0 1.0 1.0 3.0 0.0 13.0 % 4.0 1.4 0.9 1.9 0.0 1.8

Male chairman Obs. 4.0 0.0 0.0 1.0 0.0 5.0 % 2.0 0.0 0.0 0.6 0.0 0.7

Female chairman Obs. 0.0 0.0 1.0 0.0 0.0 1.0 % 0.0 0.0 0.9 0.0 0.0 0.14

Elected group (5 officials) Obs. 34.0 26.0 23.0 50.0 60.0 193.0 % 17.2 35.6 21.3 31.1 30.5 26.2

All are equally influential and Obs. 152.0 46.0 83.0 107.0 137.0 525.0

participate in decision-making % 76.8 63.0 76.9 66.5 69.5 71.2

Total Obs. 198.0 73.0 108.0 161.0 197.0 737.0 % 100.0 100.0 100.0 100.0 100.0 100.0

Source: Own census data. Test for significant difference between districts: Pearson chi2(16) = 38.50, Pr < 0.001.

When asked about influence in decision-making, about 71% of the group leaders state that all members are

equally influential while 26% state that the elected group of five are more influential. Male dominance is

not stated to be strong among the group leaders (who mostly are males).

Ostrom’ DP 4: Internal monitoring system

We have already in the previous section seen that there is a system with frequent meetings and punishments

for non-participation and late arrival. This implies one form of internal monitoring related to decision-

making. In addition, we assess whether there is a monitoring system for not coming to the work activities

of the group and for late coming to such group work activities (Tables A14 and A17).

33

Table A 14. Does the group have a bylaw stipulating penalties for not coming to work activities

of the group? Raya

Azebo

Degua

Tembien

Seharti

Samre

Kilite

Awlalo

Adwa Total

No Obs. 21 2 14 7 12 56

% 10.5 2.7 12.8 4.4 6.1 7.6

Yes Obs. 180 71 95 154 186 686

% 89.6 97.3 87.2 95.6 93.9 92.4

Total Obs. 201 73 109 161 198 742

% 100 100 100 100 100 100

Source: Own census data. Test for significant difference between districts: Pearson chi2(4) = 12.21, Pr< 0.016.

Table A 15. Have any group members been punished for not coming to group work activities? Raya

Azebo

Degua

Tembien

Seharti

Samre

Kilite

Awlalo

Adwa Total

No Obs. 146.0 38.0 84.0 141.0 139.0 548.0

% 72.6 52.1 77.1 87.6 70.2 73.9

Yes Obs. 55.0 35.0 25.0 20.0 59.0 194.0

% 27.4 47.9 22.9 12.4 29.8 26.1

If yes, how many punished? % 4.4 4.9 3.8 5.1 3.8 4.3

Total Obs. 201.0 73.0 109.0 161.0 198.0 742.0

% 100.0 100.0 100.0 100.0 100.0 100.0

Source: Own census data. Test for significant difference between districts: Pearson chi2(4) = 35.77, Pr < 0.0001.

Table A 14 shows that 92.6% of the groups have bylaw stipulating penalties for not coming to group

work activities. Table A 15 shows that 26.1% of the groups have members that have been punished for

not coming to group work activities.

Table A 16. Does group have bylaw for penalties for late coming to group work activities? Raya

Azebo

Degua

Tembien

Seharti

Samre

Kilite

Awlalo

Adwa Total

No No 33.0 9.0 38.0 37.0 32.0 149.0

% 16.4 12.3 34.9 23.0 16.2 20.1

Yes No 168.0 64.0 71.0 124.0 166.0 593.0

% 83.6 87.7 65.1 77.0 83.8 79.9

Total No 201.0 73.0 109.0 161.0 198.0 742.0

% 100.0 100.0 100.0 100.0 100.0 100.0

Source: Own census data. Test for significant difference between districts: Pearson chi2(4) = 21.99, Pr < 0.0001.

Table A 16 shows that 79.9% of the groups have bylaws stipulating penalties for late coming to group

work activities and Table A 17 shows that 21.7% of the youth groups have punished some members for

late arrival to group work activities.

34

Table A 17. Have any group members been punished for late arrival to group work activities? Raya

Azebo

Degua

Tembien

Seharti

Samre

Kilite

Awlalo

Adwa Total

No Obs. 148.0 44.0 91.0 141.0 157.0 581.0

% 73.6 60.3 83.5 87.6 79.3 78.3

Yes Obs. 53.0 29.0 18.0 20.0 41.0 161.0

% 26.4 39.7 16.5 12.4 20.7 21.7

If yes, how many punished? % 4.4 5.5 3.1 3.3 5.3 4.6

Total Obs. 201.0 73.0 109.0 161.0 198.0 742.0

% 100.0 100.0 100.0 100.0 100.0 100.0

Source: Own census data. Test for significant difference between districts: Pearson chi2(4) = 26.53, Pr < 0.0001.

Finally, we aggregate the extent of use of the four types of sanctions (absence from and late arrival to

meetings and to group work activities) by adding four dummy variables for whether each group has used

each of the four types of sanctions. The distribution of aggregate use of the sanctions is presented in Table

25, by district.

Table A 18. To what extent have groups penalized their group members by number of types of

violations, by district. Number of types if violations Raya

Azebo

Degua

Tembien

Seharti

Samre

Kilite

Awlalo

Adwa Total

0 Obs. 86.0 17.0 57.0 98.0 72.0 330.0

% 42.8 23.3 52.3 60.9 36.4 44.5

1 Obs. 33.0 12.0 16.0 22.0 41.0 124.0

% 16.4 16.4 14.7 13.7 20.7 16.7

2 Obs. 35.0 12.0 16.0 21.0 38.0 122.0

% 17.4 16.4 14.7 13.0 19.2 16.4

3 Obs. 27.0 14.0 14.0 15.0 24.0 94.0

% 13.4 19.2 12.8 9.3 12.1 12.7

4 Obs. 20.0 18.0 6.0 5.0 23.0 72.0

% 10.0 24.7 5.5 3.1 11.6 9.7

Total Obs. 201.0 73.0 109.0 161.0 198.0 742.0

% 100.0 100.0 100.0 100.0 100.0 100.0

Source: Own census data. Test for significant difference between districts: Pearson chi2(16) = 57.88, Pr < 0.0001.

Table A 18 shows that 44.5% of the groups have used none of the sanctions, 16.7% have used one type of

sanctions, 16.4% have used two types of sanctions, 12.7% three types, and 9.7% have used all four types

of sanctions. This does not mean that each of the sanctions are graduated, however, which is the focus of

the next of Ostrom’s DPs. It shows that monitoring is practiced and punishment for violation of bylaws is

implemented quite widely within the groups. This probably implies that the bylaws are very important for

compliance with the group bylaws.

35

Ostrom’s DP 5: Graduated sanctions

We have assessed the existence of bylaws within groups for not coming to and late arrival to group meetings

and to joint group work activities and the extent to which there are specified graduated sanctions as part of

the bylaws for non-compliance with these. We found that monetary and non-monetary sanctions were

commonly used by groups that have bylaws related to participation in and late arrival to group meetings

and work activities. The non-monetary sanctions varied from simple warnings, to “last warning”, dismissal

from the group, and having to work extra for the group as punishment. The monetary penalties were in most

cases increasing with the number of violations but were in fewer cases the same amount regardless whether

the violation was the first, second or third time. Some had a mixture of non-monetary and monetary

sanctions. For example, there could be only a warning first time but then a monetary sanction the second

and third times. Another common practice was to give a monetary sanction the first two times and then

dismissal the third time. There were also mixed penalties such as combinations of monetary and additional

work obligations or a monetary sanction combined with last warning. There were also a lot of variation in

the monetary amounts that had to be paid across groups and the variation from first to second and third

violation by the same person.

Table A 19 gives an overview of the presence of bylaws for the four types of violations, and whether there

are graduated sanctions in place related to these bylaws.

Table A 19. The existence of bylaws and graduated sanctions for violations of the bylaws Type of behavior Bylaw specifying

penalty for

behavior, % of all

groups

Graduated

sanctions, % of

all groups

Graduated

sanctions, % of

those with

bylaw

Not showing up for group meetings 92.3 81.5 88.3

Late coming to group meetings 84.9 62.4 73.5

Not showing up for group joint work

activities

92.5 72.0 77.8

Late coming to group joint work activities 79.9 58.2 72.8

Source: Own census data.

Table A 19 shows that it is the rule rather than the exception to have bylaws with penalties for violations

(80-93%), while 51-82% of all groups have graduated sanctions for the four types of violations, giving 72-

88% of the bylaws with sanctions applying graduated sanctions.

36

Table A 20. Number of types of graduated sanctions related to participation in group meetings

and group work activities, by district Number of types of

graduated sanctions

Raya

Azebo

Degua

Tembien

Seharti

Samre

Kilite

Awlalo

Adwa Total

0 Obs. 35.0 4.0 22.0 17.0 8.0 86.0

% 17.4 5.5 20.2 10.6 4.0 11.6

1 Obs. 23.0 8.0 6.0 18.0 14.0 69.0

% 11.4 11.0 5.5 11.2 7.1 9.3

2 Obs. 37.0 11.0 25.0 24.0 32.0 129.0

% 18.4 15.1 22.9 14.9 16.2 17.4

3 Obs. 17.0 18.0 16.0 30.0 44.0 125.0

% 8.5 24.7 14.7 18.6 22.2 16.9

4 Obs. 89.0 32.0 40.0 72.0 100.0 333.0

% 44.3 43.8 36.7 44.7 50.5 44.9

Total Obs. 201.0 73.0 109.0 161.0 198.0 742.0

% 100.0 100.0 100.0 100.0 100.0 100.0

Source: Own census data. Test for significant difference between districts: Pearson chi2(16) = 50.75, Pr < 0.0001.

We also assessed the extent to which there is variation in existence of such bylaws and graduated sanctions

across the five districts. We created a sum for the four dummies for the graduated sanctions to assess the

aggregate distribution of the four types across groups and districts.

We can conclude that graduated sanctions are considered an important tool for ensuring compliance with

the group bylaws as 88.4% of the groups have at least one type of graduated sanction in relation to

participation in group meetings and group work activities.

Ostrom’s DP 6: Conflict resolution system

Table A 21 shows that about 25% of the youth groups have experienced at least one dispute and about

12% have experienced a serious dispute. Disputes between the group and some outsiders were more likely

to be of the serious type than disputes within the group.

37

Table A 21. Seriousness of dispute versus who were involved in the dispute

The group

versus some

outsiders

Some group

members

versus

outsiders

Internal

dispute within

the group

No

response Total

No disputes Obs. 0.0 0.0 0.0 554.0 554.0

% 0.0 0.0 0.0 99.6 74.7

Some serious disputes Obs. 41.0 2.0 47.0 0.0 90.0

% 64.1 50.0 39.8 0.0 12.1

Some less serious

disputes

Obs. 23.0 2.0 71.0 2.0 98.0

% 35.9 50.0 60.2 0.4 13.2

Total Obs. 64.0 4.0 118.0 556.0 742.0

% 100.0 100.0 100.0 100.0 100.0 Source: Own census data.

Table A 22 shows that 82.9% of the disputes within the group were resolved within the group itself while

only 19.2% of the disputes between the group and outsiders were resolved among the parties themselves.

As much as 39.7% of the disputes between the groups and outsiders were still unresolved at the time of the

interview while only 3.4% of the within-group disputes were still unresolved.

Table A 22. How were disputes resolved versus who were involved in the dispute How have the disputes been resolved? The group

versus some

outsiders

Some group

members

versus

outsiders

Internal

dispute

within the

group

Total

Solved among the parties themselves Obs. 12.0 1.0 97.0 110.0

% 19.1 25.0 82.9 59.8

Resolved with help of local conflict

mediators

Obs. 4.0 0.0 0.0 4.0

% 6.4 0.0 0.0 2.2

Resolved with help from the Land

Administration Committee

Obs. 5.0 1.0 1.0 7.0

% 7.9 25.0 0.9 3.8

Resolved with help from tabia officials Obs. 11.0 1.0 12.0 24.0

% 17.5 25.0 10.3 13.0

Resolved with help from woreda officials Obs. 6.0 1.0 3.0 10.0

% 9.5 25.0 2.6 5.4

Unresolved Obs. 25.0 0.0 4.0 29.0

% 39.7 0.0 3.4 15.8

Total Obs. 63.0 4.0 117.0 184.0

% 100.0 100.0 100.0 100.0 Source: Own census data.

We also see from Table A 23 that there was a high level of satisfaction (96.6% were satisfied) with how

within-group disputes had been resolved while 23.4% were not satisfied with how the disputes between the

38

group and some outsiders had been resolved. We can conclude that local informal conflict resolution works

well for within-group disputes and only in few cases have they had to get help from community (tabia) or

district (woreda) officials.

Table A 23. Satisfaction with dispute resolution versus who were involved in the dispute Who were involved Satisfied with how the dispute was resolved

No Yes Total

The group versus some outsiders Obs. 15.0 49.0 64.0

Row % 23.4 76.6 100.0

Col % 75.0 29.5 34.4

Some group members versus outsiders Obs. 1.0 3.0 4.0

Row % 25.0 75.0 100.0

Col % 5.0 1.8 2.15

Internal dispute within the group Obs. 4.0 114.0 118.0

Row % 3.4 96.6 100.0

Col % 20.0 68.7 63.44

Total Obs. 20.0 166.0 186.0

Row % 10.8 89.2 100.0

Col % 100.0 100.0 100.0

Source: Own census data.

Ostrom’s DP 7: Institutional recognition: Minimum rights to organize

The youth groups in Tigray have been formed through a formal institutional arrangement as primary

cooperatives and therefore fall under the law of cooperatives. The group is as such an official registered

unit of business with an exclusive number of members, that is provided a demarcated resource it is

responsible for managing in a sustainable way (except for mining where it has a time-limited right to extract

the non-renewable resource). The group has to have a business plan, have an officially elected leadership

team, can apply for joint loans, and to be audited by the cooperative organization. Still, the groups are given

the authority to organize themselves and have their own bylaws. The groups are monitored and provision

of loans and legal documents for their land entitlement depends on good performance and compliance with

the rules for resource management (taking good care of rehabilitated lands). We therefore conclude that the

groups get the minimum rights to organize themselves (Design Principle 7).

Ostrom’s DP 8: Nested enterprises: Common Pool Resource that is part of a larger system

The allocation of rehabilitated communal lands to youth groups in Tigray is clearly a part of a larger plan

for sustainable land management at community, district and regional levels. Parts of the areas allocated to

youth groups are area exclosures that the communities at an earlier stage agreed to protect from resource

extraction while they also invested in the conservation of the areas by building various forms on soil and

39

water conservation structures. Some areas have also been planted with trees. Other areas are rehabilitated

gulley areas. The rehabilitation has thus typically involved labor-intensive investments where labor has

been mobilized through compulsory labor provided by community members, through Food-for-work and

Cash-for-work activities under the Productive Safety Net Program (PSNP) or other food security and

sustainable land management programs.

The community members that have invested a lot of their labor into these communal lands expect also to

get some of the benefits from these investments. It is therefore not obvious that they agree that these lands

are given to youth groups unless they have youth themselves that are included among the beneficiaries.

Therefore, all the youth that have been allocated a rehabilitated area come from the same community as the

area is located. There is variation from community to community in terms of how much land is suitable for

such distribution and that communities also agree to allocate to youth groups. However, the number of

landless youth in each community is growing by the day and community motivation to allocate land to such

youth has therefore also increased. The allocation extends the livelihood options for youth who are more

likely to remain within the community such that outmigration reduces. Most of the youth being part of such

a youth group continue to live with their own family unless they have married and have been able to get a

plot to build their own house if they do not stay with their family. The youth allocated such land are landless

and will therefore not be the ones taking over the land from their parents (landlessness is one of the criteria

for being able to join a youth group).

Appendix 2A. Additional statistical tests and regressions

Correlations between DP indicators

Table A 24. Correlation coefficients for the DP variables DP1 DP2 DP3 DP3b DP4 DP5 DP6

DP1: Encroachment control

system in place

1

DP2: Bylaw regulating sharing

arrangements

0.3105 1

DP3: Bylaw regulating

frequency of meetings

0.2522 0.3388 1

DP3b: All members involved in

decisions

-0.0738 -0.0157 0.0789 1

DP4: Number of bylaws

regulating management

0.2235 0.3425 0.7085 0.167 1

DP5: Number of graduated

sanctions

0.0765 0.2015 0.4205 0.1707 0.6144 1

DP6: Satisfactory conflict

resolution system

-0.0103 0.0199 0.0314 0.0447 0.0223 -0.0035 1

Source: Own census data.

40

Table A 25. Correlations between DPs and control variables: Assessment of endogeneity ODP1 DP2 DP3 DP3b DP4 DP5 DP6

Establishment year & month 0.004 0.001 -0.008d 0.018 -0.017 -0.015 -0.001

Initial member number 0.000 0.000c 0.000 0.002d -0.003 -0.001 0.000

Female share in group at start 0.126a 0.006 0.071d -0.051 0.551b 0.744b 0.038

Gender of group leader, 1=Female -0.047 0.011 -0.007 0.046 0.083 -0.060 -0.017

Distance to road, km 0.001 0.004c -0.001 -0.008 -0.018 -0.032 -0.001

Distance to market, km -0.001 -0.001 -0.001 -0.006d 0.000 0.003 -0.001

Distance to home, min. km -0.002 -0.007 0.000 -0.010 0.011 0.085c 0.003

Input access problems, dummy 0.035d 0.002 -0.007 -0.052 -0.073 -0.108 -0.014

Baseline: No conflicts 0.000 0.000 0.000 0.000 0.000 0.000 0.000

Serious conflict in group, dummy 0.009 0.014 0.035d 0.042 0.178d 0.188 -0.215a

Less serious conflict in group, dummy -0.001 -0.013 0.008 0.103c -0.118 -0.094 0.008

Area allocated per person in group -0.018 0.019 0.006 0.136d 0.220 0.114 0.007

Number of returning migrants 0.004d 0.000 0.001 -0.007 0.002 -0.027c 0.001

Change in key group positions, dummy 0.059a 0.019 0.000 0.015 0.023 -0.115 0.014

Self-selection of group members -0.036b -0.005 -0.028c 0.400a -0.075 -0.222d 0.000

District dummies: Base=Raya Azebo 0.000 0.000 0.000 0.000 0.000 0.000 0.000

Degua Tembien 0.037 0.088a 0.070b -0.032 0.307c 0.207 0.023

Seharti Samre 0.059 0.085a 0.025 0.034 -0.230 -0.160 0.047b

Kilite Awlalo 0.128a 0.047d 0.098b 0.044 0.224d 0.244 0.024

Adwa 0.103a 0.090a 0.037 -0.065 0.125 0.384c -0.017

Activity dummies: Base=Animal rearing 0.000 0.000 0.000 0.000 0.000 0.000 0.000

Beekeeping -0.031 0.001 -0.004 -0.007 0.012 0.047 -0.024

Forest 0.014 -0.015 -0.027 -0.031 -0.171 -0.392 -0.007

Irrigation/Horticult -0.054 0.010 -0.016 -0.022 0.101 -0.047 -0.003

Mining -0.036 0.050 -0.050 0.007 -0.052 -0.221 -0.013

Other 0.028 0.020 0.015 0.417a -0.158 0.090 -0.030

Constant -6.996 -1.035 16.103d -35.296 38.403 32.979 2.919

F value 1.995 0.826 1.546 5.386 2.137 2.712 1.293

Prob > F 0.004 0.699 0.050 0.000 0.002 0.000 0.162

Number of observations 727 727 727 727 726 727 727

R-squared 0.093 0.060 0.051 0.154 0.057 0.066 0.203 Source: Own census data. OLS models for each DP. The table shows marginal effects. Significance levels: d < 0.10, c < 0.05, b < 0.01, a < 0.001.

41

Conflict experience and the DPs

Table A 26 shows the extent of exposure to less serious and more serious disputes among the youth groups.

The results from ordered probit models and multinomial logit models for the DPs’ relationship with less

serious and more serious disputes are presented in Table A 27. Table A 27 demonstrates that the pre-

determined DPs to a small extent are associated with the occurrence of less serious or serious disputes. For

DP3b (all members involved in decisions) we see a positive correlation (significant at 5% level) with the

occurrence of less serious disputes. This could be related to group dynamics. If many members want to

influence decisions this could lead to internal disputes. For DP6 (satisfactory conflict resolution system in

place) there is a strong negative correlation with serious dispute (significant at 0.1% level in the multinomial

logit model while DP was insignificant and with a positive sign in the less serious conflict outcome of the

multinomial model). It is possible that exposure to serious conflicts have exposed awareness of limitations

of the conflict resolution system such that the causality goes from serious conflict to DP6 which is a

perception variable. Many of the serious disputes are related to external agents and could not be handled

adequately by the local conflict resolution system. Many of these disputes are still unresolved.

Table A 26. The extent of conflict experiences of youth groups by 2016

Has the group experienced any conflicts since its establishment?

Variable code Response Freq. Percent Cum.

0 No conflict 554 74.66 74.66

1 Less serious dispute 98 13.21 87.87

2 Serious dispute 90 12.13 100.00 Total 742 100.00

Source: Own census data.

We briefly inspect the control variables in Table A 27. There are significant variations across districts in

the extent of less serious and serious disputes but no significant differences between types of main activities

of the youth groups. Self-selection of youth group members was associated with lower probability of

(serious) disputes. Less serious disputes were also more likely to groups established earlier rather than later.

This could be because the aggregate probability of conflicts occurring may increase over time but could

also be due to less experience with formation of groups at an early stage of this large-scale social

experiment. Serious conflicts were also positively associated with the area allocated per youth. If the value

of the land is increasing with the area, there may be more conflicts and competition over a larger land

resource.

42

Table A 27. Ordered probit and Multinomial logit models for factors associated with groups

having had serious and less serious disputes Ordered Multinomial logit model

probit Less serious

dispute

Serious

dispute

DP1: Encroachment control system in place 0.114 0.432 -0.062

DP2: Bylaw regulating sharing arrangements 0.106 -0.497 0.785

DP3: Bylaw regulating frequency of meetings 0.477 1.167 1.021

DP3b: All members involved in decisions 0.299c 0.753c 0.307

DP4: Number of bylaws regulating management 0.012 -0.276 0.202

DP5: Number of graduated sanctions -0.022 -0.038 -0.030

DP6: Satisfactory conflict resolution system -2.573a 14.818 -4.476a

Establishment year and month -0.072c -0.207b -0.099

Number of initial members in group 0.003 0.002 0.006

Gender youth group leader, female=1 dummy -0.158 -0.875d 0.010

Distance to road, km -0.056c -0.072 -0.128d

Distance to market, km 0.014 0.042d 0.016

Distance to home, min. km 0.017 0.026 0.028

Input access problems, dummy 0.173 0.231 0.343

Share of females in group at start -0.358 0.084 -1.081

Land area per person allocated to group, ha 0.570c 0.864 1.391c

Number or returning migrants in group 0.002 -0.013 0.001

Self-selection of group members, dummy -0.324c -0.125 -0.922c

District dummies: Base=Raya Azebo 0.000 0.000 0.000

Degua Tembien -0.143 1.726b -1.559c

Seharti Samre 0.116 1.104d -0.131

Kilite Awlalo -0.278 0.553 -1.059c

Adwa 0.295d 2.140a -0.296

Activity dummies: Base=Animal rearing 0.000 0.000 0.000

Beekeeping -0.047 0.095 -0.150

Forest 0.013 0.359 -0.304

Irrigation/Horticulture 0.017 0.319 -0.182

Mining -0.092 0.581 -0.449

Other -0.171 1.429 -13.377

Constant 397.295 200.320

cut1 constant -146.913c cut2 constant -146.312c Wald chi2 138.883 198.088

Prob > chi2 0.000 0.000

Number of observations 726 726

Source: Own census data. Significance levels: d < 0.10, c < 0.05, b < 0.01, a < 0.001. Coefficients are relative

risk ratios in the multinomial logit.

43

Appendix 3A. Group size distribution

Figure A2. 1. Share of group members who stay in the group in 2016 versus the initial group size

with local polynomial smooth (lpoly).

Figure A2. 2. Current group size distribution in 2016 versus group size distribution when groups

started.